Skip to main content

Full text of "Hurro-Urartian Borrowings in Old Armenian"

See other formats

Hurro-Urartian Borrowings in Old Armenian 

L M. Diakonoff 



Journal of the American Oriental Society, Vol. 105, No. 4. (Oct. - Dec, 1985), pp. 

Stable URL:^^ 

Journal of the American Oriental Society is currently published by American Oriental Society. 

Your use of the JSTOR archive indicates your acceptance of JSTOR' s Terms and Conditions of Use, available at JSTOR' s Terms and Conditions of Use provides, in part, that unless you 
have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and 
you may use content in the JSTOR archive only for your personal, non-commercial use. 

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at 

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or 
printed page of such transmission. 

JSTOR is an independent not-for-profit organization dedicated to creating and preserving a digital archive of 
scholarly journals. For more information regarding JSTOR, please contact 

ThuMar 9 09:11:27 2006 



Leningrad, USSR 

This study examines a number of lexemes in Armenian which appear to have Hurro-Urartian 
etymologies. It attempts also to isolate the chronological periods during which these borrowings 
took place and to describe the linguistic changes that these words underwent in Armenian. 

Armenian is an Indo-European language which 
recalls English in that it is saturated with borrowed 
lexical material. Apart from the usual amount of 
occasional borrowings for which it is difficult to 
establish a valid historical reason, most of the borrow- 
ings correspond to certain cultural innovations in the 
Armenian society itself. In this paper we shall not dwell 
on the strata which involve Neo-Armenian. However, 
in the Old Armenian (Grabar), which up to the 19th 
century was the literary language for all Armenians, 
there are also several clearly definable strata of 
borrowings. Grabar is a literary language whose core 
was formed between the 5th and the 8th centuries a.d., 
but the stratum of borrowings which shall be discussed 
here goes back to a period before the 5th century b.c. 
Roughly, the following strata in Old Armenian can be 

(1) Middle Iranian (predominantly Parthian but 
also Middle Median and especially Middle Persian) is 
responsible for the greatest number of borrowings. 
They denote abstract notions, terms of the feudal 
society, but the strata in question also contain a 
considerable number of words denoting objects of 
daily life; they have probably ousted older lexemes 
with a similar semantic of a Common Indo-European 
origin. Note that there are few certain Old Iranian 
borrowings, if any; 

(2) The Syriac Aramaic dialect, and to a lesser 
degree Greek (partly also through the mediation of 
Syriac) are responsible for words referring to religious 
and ecclesiastical matters, and also to writing and 
literary activities etc.; 

(3) A separate stratum consists of borrowings from 
an earlier Aramaic dialect (possibly connected with the 
resettling to Armenia of city-dwellers of Aramaic and 
Jewish origin under Tigranes II and Artavazdes I in 
the 1st century B.C., but also with the Aramaic traders, 
scribes etc. of the Hellenistic and Achaemenian pe- 
riods). This stratum is responsible for terms connected 
with trade and traders, names of wares which were 

in circulation, some implements etc. These words 
are mostly easily discernible because they preserve a 
number of early North Aramaic peculiarities and mostly 
lack the typical final Aramaic -a (> Arm. -ay) which 
at that epoch was still an article in Aramaic. A 
number of originally Akkadian words have passed into 
Armenian probably via Aramaic mediation; 

(4) Below these strata lies a stratum whose origin 
was until lately undetermined, and which was usually 
referred to as the "Caucasian substratum". A consider- 
able number of words have been identified (mainly by 
Hr. Adjarian) as originating from certain Kartvelian 
(Georgian) dialects; some were thought to be akin to 
Udi (a southern dialect of the Lezghian group in the 
North-Eastern Caucasian linguistic family), or as 
originating from the languages of Asia Minor and the 
Aegean world. However, hundreds of words are hither- 
to unexplained, even tentatively. 

Since the Armenian nation, as seems to follow from 
written sources, has been formed in the territories 
inhabited in the 2nd millennium B.C. by Luwian, 
Hurrian and Urartian speakers, a search for Luwian, 
Hurrian and Urartian (or Hurro-Urartian) 1 etymo- 
logies applicable to Armenian words, might be reward- 
ing. Much work has been done in this direction 
by a number of authors, 2 but it is only in the last 
few decades that our knowledge of the languages in 
question has increased sufficiently, and now this 
problem has been tackled by J. A. C. Greppin, 3 G. B. 

Apart from the written Hurrian and Urartian languages 
there must have existed a number of intermediate dialects 
which had not been committed to writing. 

I am thinking of the well-known works of P. de Lagarde, 
J. Markwart, N. G. Adontz, Hr. Adjaryan, G. A. Kapantsyan 
and others. 

J. A. C. Greppin, Hittite -za and Armenian z- and the 
"Theory of Armeno- Hittite borrowings," Journal of Indo- 
European Studies 3 (1975), id., Armeno- Luwica, Revue des 



Journal of the American Oriental Society 105.4 (1985) 

Djahukyan, 4 N. Mekertchyan, 5 N. V. Harouthiounyan 
(toponymies), and the author of the present paper. 6 
In the following we will try to present a list of more 
or less dependable results, hitherto achieved: 7 


1. aiaxin 'servant girl, slave girl'< Hurr. al(l)a(e)h- 
he/inne % 'keeper (male or female) of that which pertains 
to the lord of the house/ family' i.e., of household (or 
temple) stores, mostly of food (cf. Akkad. ala(h)hinnu 
do., 9 NAss. lahinu, lahinatu do., Mishn. Hebrew 
hhena, Aram, hhenta 'concubine'). The Hurrian form 
should be analyzed as *all-ae 'lady (of the house)' (but 
dialectally 'lord of the house' is also possible), 10 -hhe 

etudes armeniennes 13 (1978/79); id. "Armenian laxur and 
Hittite lahhur," ibid., 15 (1981); id., "The Anatolian substrata 
in Armenian — an Interim Report," Annals of Armenian 
Linguistics (1982); id., "Luwian Elements in Armenian," Hin 
Arevelk c — Drevnij Vostok 3, Erevan, 1978, pp. 115-26, 
267-68, et al. 

4 G. B. Dzhaukjan (Djahukyan), Ocerkipo istorii dopis'men- 
nogo perioda armjanskogo jazyka [Studies in the Pre- Literate 
Period of the Armenian Language], Erevan, 1967; id., 
Khajasskij jazyk i ego otnosenie k indoevr ope j skim jazykam 
[Haiasan in its Relation to Indo-European], Erevan 1964; id., 
'Genetic Coincidences in the Vocabulary of Armenian and 
Hittito-Luwian," Patma-banasirakan handes 24 [39] (1967) 
(in Armenian); id., "Novye urartsko-indoevropejskie paralleli 
[New Parallels between Urartianand IE]," Izvestija Akademii 
nauk Arm. SSR, 3 (1965); id., "Hittito-Luwian Elements in 
the Armenian Vocabulary," Vestnik Erevanskogo Universiteta 
2 (1967) (in Armenian). 

5 N. Mekertchyan, "Verbal Reduplication in Hittite and 
Armenian," Hin Aravelk c - Drevnij Vostok 2, Erevan, 1976 
(Russian, English summary: The author envisages borrowings 
from Hittite, but they are more likely to be from Luwian or 
Hittito-Luwian); id., "Substratum Names of Plants in 
Armenian," ibid., 4, 1983 (Russian, English summary), to be 
used with caution. 

I. M. Diakonoff, Hurrisch und Urartaisch, Munich 1971. 

Where the author of the etymology is not mentioned, it 
means that the etymology has long been accepted in the 
special literature; some of the anonymous etymologies have 
been proposed by the author of the present paper. 
8 PI. a-la-ah-he-en-ne-na HSS 15 211:28. 

CAD, AHw s.v. Note that von Soden's translation, 'miller', 
is wrong. In the following all Akkadian glosses are taken from 
these two dictionaries, unless another source is specifically 

10 Hurro-Urartian had no genders, and the form all-ae is 
derived from the root *all-, Urart. al- 'to be lord of, to rule' 
(cf. e.g., Urart. al-usd 'ruler'). The root is Common PEC. 

possessive adjective suffix, 4- -(i)nne suffix of names of 
professions and similar. Note that Hurro-Urartian -ae- 
has a tendency to develop to -e-, -e in Hurrian, but to 
-a(-) in Urartian. Hence the borrowing may be from 
Urartian or a dialect proximate to literary Urartian. 
Note also that the Armenian reflex -/ does not 
necessarily point to a reduplicated *-//- in the original 
(as I formerly thought); the spelling / or // depends 
upon whether the Hurro-Urartian phoneme is from the 
Proto-Eastern-Caucasian (PEC) lateral sibilant, or 
from the PEC / or l\. It is well known that, different 
from Hurrian, Urartian had no long or reduplicated 
consonants, at least in spelling. 

2. aix 'household, household property' < Hurro- 
Urart. *all-ae-hha, cf. No. 1. 

3. astern 'I marry' < Hurrian aste 'wife' (Kapan- 

4. ear ay 'slave' > *car(r)a, the Hurrian variant being 
sarre < *carr-ae 'live booty, captives', Akkad. sallatu 
(spelled in an Akkadian lexical text as sar-ri/e) n Note 
that Hurrian s = Urart. s n was an affricate, /c/ or /£/; 
all 'emphatics' yield simple consonants in Armenian, 
while the non-emphatics yield aspirated consonants. 13 
Of course, the Akkadian spelling could also allow of a 
reading sar-ri/e, but Hurrian sarri is another word 
meaning 'king' (borrowed from Akkadian). 


5. agur 'burned brick' < Akkad. agurru < Sum. al- 
ii r-(r) a do., probably through Hurrian mediation. 
In Aramaic only the form agora (borrowed from 
Akkadian) is attested; an earlier Aramaic dialect might 
have preserved a form *agurr-a or, without the article, 
*agur but, as Professor Anahit Perikhanian tells me, 
these forms would yield in Armenian *aguray (from 
Syriac) or *agur (from earlier Aramaic), not agur. The 
same must be true in the case of an immediate borrow- 
ing from Akkadian. 

6. anag 'tin'< Hurr. *anagi < Akkad. a(n)naku do. 
(< Sum. a n-(n) a. A Hurrian mediation is here 

E. Laroche, Glossaire de la langue hourrite (sub sarri), 
Paris, 1980, p. 217. The Hurrian glosses, if not noted other- 
wise, are taken from this publication. 

12 Cf. Hurr. pis- 'to rejoice'= Urart. pis-do., et al. (<PEC 
*c, *c). 

13 See I. M. Diakonoff, S. A. Starostin, Hurro-Urartian as 
an Eastern Caucasian Language, mss (in press). 

14 The Sumerian has the readings a n-n a and 
n a g-(g) a, n i g-(g) i, both of which have been borrowed, 
cf. Akkad. ainjna/ku'lead, tin', Hebrew °anak 'plumb', Aram. 
°an(nd)k-a 'tin', Sanskr. naga- 'tin'(M. Mayrhofer, Kurzgefas- 
stes etymologisches Worterbuch des Altindischen, Heidelberg, 

Diakonoff: Hurro-Urartian Borrowings in Old Armenian 


nearly a certainty, because only in Hurrian but not in 
Urartian and Aramaic is *-k- in medial position reflected 


as -g-. 

7. knik c 'seal' < Hurr. *kanikki < Akkad. kaniku 
or, according to Reiner's law, alternatively kanikku 'a 
sealed object' (document, sack, bulla, etc.). The dif- 
ference of reflexes between the initial and the medial k 
points to a Hurrian mediation. The common etymo- 
logy < Akkad. kunukku 'seal-cylinder' must be re- 

8. p c ox 'barter, exchange' < Akkad. puhu do. The 
transition of u > b could find place already in Ak- 
kadian, since Greek transcriptions of late Babylonian 
words attest it, although we do not know when this 
phenomenon actually appeared in Akkadian. But also 
in Hurro-Urartian /o/ was much more frequent than 
/u/. The word is actually attested in Hurrian: pdhi(!), 
poh-ugar- (-ugar- being a suffix of reciprocity) 'barter, 
exchange'. It probably must have existed also in 

9. t c iw "number' < Hurr. tiw- 'to say, speak', tiwd 
'word, deed' (Kapantsyan). 


10. ananux 'mint'< Hurr. *an-an-uhha or *an~an- 
uya. Actually attested are Akkad. anariihu possibly 
'mint' < Hurr. *an-an-ihha J and Hurr. an-an-isha, 

1959 sq., II, 150); it is not impossible that both Sum. 
n a g-(g) a and Sanskr. naga- might be borrowings from the 
same common source (the language of the Harappa -culture?), 
and that only a n-(n) a is genuine Sumerian. Akkad. 
a(n)naku means 'tin or lead' (cf. H. Freydank, "Fernhandel 
und Warenpreise nach einer mittelassyrischen Urkund des 12. 
Jahrhunderts v.u.Z.,"in Studies in Honor of I. M. Diakonoff, 
Warminster 1982, pp. 64, 72), and a n-(n) a is even 'iron' 
(cf. A. A. Vaiman, "Eisen in Sumer," AfO 19 (1982), S. 33-38). 
But Arm. anag cannot be etymologized from either Sansk. 
naga- or Sum. n a g-(g) a but only from Akkadian through 

It would be tempting to include in the list Arm. arlez 
{aralez, yaralez) 'spirit accompanying the dead', etymologizing 
it from Akkadian (also Hurrian?) aral{l)e 'Underworld'. 
However, the Hurro-Urartian word-formational suffix -z/", 
fairly common both in Hurrian and in Urartian, unfortunately 
occurs only in obscure glosses, and its sense is unknown; also, 
it seems to attach only to consonants. A form *arale-zzi (with 
the adjective suffix -Vzzi) would make sense, viz., 'character- 
istic of, pertaining to the Underworld', but again unfortunately 
the -zz- in this suffix (= Urart. -usd) seems to stand for an 
unvoiced sibilant affricate and hence could hardly yield -z in 

both with suffixes which alternate with -(u)hha and 
-uya and have approximately the same possessive 
meaning. The Aramaic forms with loss of the initial fl- 
are later and derived (Mekertchyan). 

1 1 . nurn (Gen. man) 'pomegranate' < *nurn V\ prob- 
ably a variant of *nurm V, which is the prototype of 
Sum. gl§ n u-u r- m a 'pomegranate', certainly bor- 
rowed (as can be inferred also from its spelling), 
obviously from the mountaineer neighbours. Hence 
Akkad. nurmu (in Nuzi also lurmu, lurinu). In Hurrian 
the only attested form in the adjective nurandi-ya 'of 
pomegranate' which is supposed to derive from an 
Assyrian *nuramtu (the Babylonian feminine form is 
nurimtu). However, the ending -(an)-di can easily be 
explained as Hurrian word-formational suffix(es). It is 
clear that the Mesopotamian terms are borrowed, 
while the terms in the mountaineer languages have a 
good chance to be original. Other Semitic languages 
yield metathetic forms (Hebrew rimmon, Arab, rum- 
man-) (Mekertchyan). 16 

12. sa/or (East, slor) 'plum(-tree)'. The Hurrian and 
Urartian words for 'plum' are not attested, but we have 
the Akkad. salluru 'plum(-tree), or 'medlar', obviously 
a borrowing from a Hurr. I sail-oral , and Akkad. Nuzi 
sennuru (cf. Sum. sennur) do.; these forms also go 
back to a Hurrian variant of the same word, /senn-ora/ 
(an alternation -n-/ /-I- is well known in Hurrian). The 
Armenian word has certainly the same origin; whether 
from Hurr. /*sall-or3/ or from an Urartian cognate, 
viz. *saluri /sal-ora/, cannot be ascertained. 

13. serkewil 'quince'. Cf. Arab. safargal< Aram. 
*sapargala < Akkad. sapu/argillu. Here is a clear case 
of metathesis, but it is not clear which form, the 
Akkadian or the Armenian, is nearer to the original, 
that might have been Hurrian or have belonged to 
another extinct Caucasian language (Mekertchyan). 

14. tuit, teit (late also tuid) 'althea'. Mekertchyan 
compares the Akkadian tuldu (from a lexical text; 

Mekertchyan thinks also that (among others) Arm. maxr 
'a coniferous ixee* and p c arp c ar 'purslane' should be regarded 
as of Akkadian or Hurrian origin. However, the medieval and 
dialectal Arm. maxr is from Pers. marx, and also Akkad. 
mehru 'pine or spruce' may be of Iranian origin. The main 
term for 'purslane' in Armenian is kockorak, later danduf, 
while all the dialectal forms like p c erp c eran, p c erp c rem etc. 
'purslane' have their origin in Pers. parpahan. P c arp Q ar 
'purslane' in Galen is due to a mistranslation. I am indebted 
for this information to Professor Anahit Perikhanian. See 
however Akkad. parparhu 'purslane', obviously from a Hurr. 
*parpar-hd, which may also be the origin of the Aram. 
parpahTna (from the Hurr. plural in -rar?)— and of Pers. 
parpahan ? 


Journal of the American Oriental Society 105.4 (1985) 

glossed ladiru ina Subari 'Hurrian for ladiru')} 1 
Akkad. tuldu might reproduce a Hurrian *tulti\ how- 
ever, from the Akkadian context it only follows that 
this was a 'medicinal plant' (Kapantsyan). 

15. uit 'Bactrian camel' < Urart. ultu do. The Urar- 
tian word is somehow connected with Akkad. utltru 
(not *udru\) do., and perhaps with Olran usOra-. There 
is an IE etymology proposed for the latter, which is not 
overwhelmingly convincing but may be correct. 18 

16. xafof Vine'. Possibly < Hurr. /*hall-o-ta/,/w-/w- 
le in the Nuzi spelling: the term is mentioned in an 
administrative account along with uhlnu 'unripe dates' 
and may quite possibly mean 'grapes' (plural -/a). 

17. xA2/or'apple(-tree)'< Hurr. (probably also Urar- 
tian, but not attested as such) hinzuri / hinj-ors / . The 
Aramaic hazziira is certainly < Akkad. hinzuru {hi- > 
ha- is typical). The Sumerian hashur 'apple-tree' is 
also from Eastern Caucasian but not Hurrian (possibly 
from Qutian which may have belonged to Western 
Lezghian languages). 

These glosses, a dozen and a half, may be regarded 
as either certain or probable. Several more have been 
proposed, but they all have some important draw- 
back — either the interpretation of the semantic is 
inexact, or the Armenian gloss is late, or it can be 
shown to have another more plausible etymology, etc. 

Additionally we may mention a few probable 
etymologies which, however, do not fall under the 
three main semantic headings above, namely: 

18. car 'tree'. This is a crux. According to Adjarian, 
it is to be etymologized from IE *g'rso-, cf. Greek 
ydpoava 'brushwood', yeppov <* yepaov 'wattle-fence', 
Old Norse kjarr 'brushwood'. An alternative etymo- 
logy, advocated by myself and others, compares Arm. 
car 'tree' to Urart. sard 'orchard', Hurr. sar-me 
(/car-(?)/ , attested in an Akkadian lexical text) 'wood', 
both < PEC *cwal\hV 'v/ood, fire-wood; big stick'. Of 
course, the derivation of the name for such a common 
object as a tree from 'orchard', 'wood' or 'big stick' 
does not sound very convincing; but neither does the 
derivation of 'tree' from 'brushwood' or 'wattle-fence'. 
Sub judice lis est. 

Ladiru or aladlru is certainly also a borrowed word, 
which can be seen from the elision of a- typical of borrowed 
but not of original Semitic words (because in the latter the 
spelling a- stands for / 3 a-/; cf. alahhinnu / / ' lahinu et al.). 

The Urartian glosses, if not noted otherwise, are from 
I. M. Diakonoff, Urart skie pis 'ma i document y [Urartian 
Letters and Documents], Moscow-Leningrad 1963; cf. also 
the glossary in G. A. Melikisvili, Urartskie klinoobraznye 
nadpisi [Urartian Cuneiform Inscriptions], Moscow, 1960, 
continued in VDI 1971, 3-4. 

19. cov 'sea' < Urart. sua (if interpreted as /co(w)s/ , 
which is quite possible) '(inland) sea'. 

20. xarxarem i destroy' < Urart. harhar- 'to be 
destroyed' ('to destroy' is Urart. harlyar-s-). The dif- 
ficulty lies in the fact that the Armenian transitive form 
seems to be connected with the Urartian intransitive. 
However, it can easily be imagined that there could 
have existed a dialectal situation in which the difference 
between Urart. harhar-, 'to be destroyed,' and Urart. 
dial. *harhar-, 'to destroy', would be expressed not by 
the suffix -s- but by the more common method of 
changing the intransitive personal morphs for the 

Even if we discard the last three items as occasional 
or doubtful borrowings, the list of such borrowings 
from Hurro-Urartian into Armenian as can be regarded 
as certain or probable remains highly interesting. It has 
a bearing on the question, whether Proto-Armenian 
was an aboriginal language, being at least as old in its 
homeland as Hurro-Urartian, or even older, 19 or it was 
introduced in the Highland over a Hurro-Urartian 
substratum in the 12th century B.C., as 1 think, and as 
many predecessors (the latest being H. A. Manandyan) 
have thought before me, suggesting even later dates. 20 

The IE kernel of Old Armenian contains all the 
necessary words denoting man, parts of the body, 
natural actions and states and also the most important 
terms for the domestic animals, except the camel. 
Of course, in the 2nd millennium B.C. the Proto- 
Armenians could not have been nomadic cattle- 
breeders. They had, no doubt, also a subsidiary 
agriculture, which is attested, e.g., by the IE words for 
'barley' (gari) and the 'plough' (arawr) in Old Armenian. 
But they had to borrow from the Hurro-Urartians the 
most necessary terms of a settled agricultural early 
class civilization (such as 'slave', 'slave-girl', 'burned 
brick', 'tin', 'seal'), as well as words for local animals 
and plants ('camel', 'apple', 'plum' or 'medlar', 
'quince'/?/). The only possible conclusion is, that the 
immigration of the Proto-Armenian speaking tribes 
postdated the settlement of the Hurro-Urartians in the 

F. Kortlandt 21 has suggested a relative chronology 
for the various phonetic changes which occurred at the 

As suggested by Vyacheslav Vs. Ivanov, Th. V. Gam- 
krelidze, G. Klychkov, O. Shirokov and a number of other 

See I. M. Diakonoff, The Prehistory of the Armenian 
Nation, Delmar, N.Y. 1985. 

F. Kortlandt, "On the relative chronology of Armenian 
sound changes," First International Conference of Armenian 
Linguistics: Proceedings, Delmar, N.Y. 1980, pp. 97-106; 

Diakonoff: Hurro-Urartian Borrowings in Old Armenian 


different stages of the development of Proto-Armenian. 
In reconstructing these stages, it is necessary to take 
into consideration that at the later stages there must 
have existed an Armenian- Urartian bilingualism, which 
must have influenced the process of the changes. 

The changes from PIE to Proto-Armenian and those 
inside Proto-Armenian which antedated the contact 
with Hurro-Urartian can be inferred from the above 
list of borrowings of lexical material from Hurro- 
Urartian into Old Armenian, and also from toponyms 
inherited by the Armenians from the Urartians, such as 
Abiliana /Abiliana/, Arm. Abetean, Alzi- /Aldzi-/, 
Arm. Aij-ni-k c , Biaine-h /Viane-b/, Arm. Van, 
Gaura-ha, Arm. Gawre-k c , Giarnia-, Arm. Garni, 
Halitu-, Arm. XaH-k c , Qutumu- /Qot D omV/, Arm. 
Kotom, Suluqu- /Culuk D u-/, Arm. Cfuk, Supa- 
/Copa-/, Arm. Cop c -k c , Seseti, Arm. Sawsat c (l), 
Tuarasd /T D uaracV/, Arm. Tuaraca-tap c , Tuspa 
/T D osp 3 a/, Arm. Tosp, Zabahae /Javaha/, Arm. Javax- 
k c . 22 These toponyms, as well as the borrowings from 
Hurro-Urartian into Armenian of appellatives quoted 
above, show beyond any doubt that the contact of the 
speakers of Proto-Armenian with the Hurro-Urartians 
took place after the moment when in Proto-Armenian 
(and also in Phrygian and Thracian) the shift of IE 
voiced, voiced aspirated and unvoiced to unvoiced 
(glottalized?), voiced, and unvoiced aspirated had taken 
place and ceased to be productive. The same can 
probably be said of the change from *w> g (we have 
only a case of retainment of the Urartian /v-/, which 
may not be the same thing); and also of the change of 
*s- to zero (also shared with Phrygian). 23 

Note also two cases of a presumable reproduction of 
the Armenian plural in -k c as -ki in Urartian: Mus-ki- 
4 Proto-Armenians\ later also 'Phrygians, Phrygia', 
cf. Akkad. Muski, Hebrew Masak (more correctly 
Mosak, M6oo% in the Septuagint), 24 the same sense; 
Tumis-ki /T 3 omis-ki/ = Greek Tomisa (plural!). 

According to Kortlandt, the shift of the stops and 
the assibbilation of *k' to *s' (of which there are 
already no traces in the above material), the develop- 
ment of *s to *h (although as yet not to zero) and the 

through the kindness of A. Perikhanian I was also able to 
consult Kortlandt's ms on Proto-Armenian case endings. 
Note that the Urartian graphemic b stands for /b/ and 

fvf,p for I pi and /p/,ifor /s/ and (more seldom) for fs/ ,s 
for /c/ and /£/,sfor /c/ and /£/,zfor /dz/ and /J/,and for 
/z/ (?), u for /o/ and more seldom for /u/ and /w/, etc. 

Thus Garni, not * Kami or * Carni, agur, not *akur, salor, 
not *alor (cf. IE sal 'salt'> Arm. ai, etc.) 

The stem is *mus-, cf. Muaia, Urart. Musa /Musa(?)/, 
Luw. Hier. Musa- and Muska-. 

palatalization of the velar stops before front vowels 
ceased to be productive at stages 1 to 8, which means 
that the contacts with Hurro-Urartian postdated 
stage 8. Kortlandt dates also the development of IE *o, 
*e and *-on, *-en to (-)w, (-)/ rather early, too, which is 
corroborated by the fact that a similar development 
occurs also in Old Phrygian (8th-4th centuries B.C.). 
The development of *w to *y w to g, and the develop- 
ment of *h w (from *sw) to *x and of final -s in nom. 
plural also to *-x he dates to the rather late stage 1 1, 
which, according to him, postdates the ienition' of *p, 
*/, *k, *k w to *cp, *0, *x, *X W ( at stage 10, these phonemes 
later becoming/? , / c , k c at stage 19), but to */before r. 
This hypothesis explains a number of later phonetic 
phenomena; however some doubt remains. Thus, for 
one thing, a development from *cp, *0, *xtop c ,t c ,k c can 
hardly be regarded as 'a natural type of development'. 
But it could have taken place in the period of Proto- 
Armenian — Urartian bilingualism, since Urartian had 
no fricative phonemes of the type of (p, 9 (Hurro- 
Urartian / developed into v-, -w in Urartian). And 
if our interpretation of the terms Muski, Tumiski 
is correct, that would mean that the change from *-s 
to *-x to *-k c must have occurred considerably earlier, 
also antedating the Proto-Armenian-Hurro-Urartian 
contacts: Urartian h {— Arm. x) did not develop 
into k c in Armenian, cf. Halitu > Xafti-k c , Zabahae > 
Javax-k c . Is it possible that the 'Ienition' of */?, */, *k 
occurred only in certain phonotactic conditions, and 
not as a general rule? 25 

25 Thus, it is difficult to explain the development of the IE 
plural ending in *-s to -k c without postulating an intermediate 
*-x, or to explain the genitive hawr of hayr 'father' (< IE 
pater) without postulating such intermediate forms as *hat c ir, 
gen. *ha9ros > *hafroh >*hawrV. But the 'Ienition' may not 
have occurred, e.g., in initial positions except in the case of 
Arm. x- < IE *kH, which development Kortlandt relegates to 
stage 20 (after the apocope). The case of x may have been 
strengthened because of the fact that h (= x) was a very 
frequent phoneme in Urartian and Hurrian. The Urartian 
phonetic structure must have had a considerable influence on 
the development of the Armenian phonetic structure at the 
later stages of Proto-Armenian. Note the emergence of a 
number of phonemes unknown to PIE — not only of such as 
were due to spontaneous palatalization, as c D , ], c, /, but also 
such as were quite foreign to Indo-European, e.g., (glottalized) 
c or x: if there had existed a stage of 'Ienition', it was anyway 
shortlived, neither q> or surviving: they were not supported 
by the Urartian phonetics; but x (= Urart. h) did survive. 
Kortlandt suggests the development of Arm. ac from *aug(e) 
and places this development at stage 2. However, it is hard to 
imagine that a separate phoneme would develop at so early a 
period not supported by other occurrences except that one 


Journal of the American Oriental Society 105.4 (1985) 

This problem is connected with the question of 
the origin of the Armenian self-denomination hayo-, 
nom. plural hay-k c \ if the present author's contention 
(earlier brought forward by P. Jensen and P. Tashyan) 
that hayo- should be derived from Proto-Armenian 
* Hat c io(s) < Urart. Hati- 'name of countries west of 
the Euphrates and specifically of Melitene, and of their 
inhabitants', is correct, then the contact between 
Proto-Armenians and Urartians must have happened 
before stage 10 or after stage 19, or else the idea of 
'lenition' itself should be abandoned. This means that 
*p, */, *k developed directly into p c , / c , k c , except for 
positions before -r and perhaps some other specific 
conditions. Naturally, however, I may be wrong. 

The terminus ante quern for the beginning of a 
Proto-Armenian — Urartian bilingualism is Kortlandt's 
stage 16, to which he dates the apocope, or loss of the 
final vowel in Armenian. The apocope was obviously 
due to a change from a free tonal accent of the Greek 
or Vedic type which certainly still existed in Phrygian 
and probably in Proto-Armenian, to a fixed strong 
expiratory stress on the penultimate syllable. This was 
the type of stress typical of Urartian. The apocope led 
to a complete restructuring of the Armenian inflection; 
this, then, is also the result of Hurro-Urartian influence. 

Another important influence of Hurro-Urartian on 
Armenian is, no doubt, the loss of genders which were 
completely unknown to the former. According to 
Kortlandt, the gender distinction was lost in Proto- 
Armenian (save for a few relics in the pronoun) before 
the apocope. Hence the date of the beginning of the 
bilingualism must date also from a stage earlier than 
stage 16. 

As mentioned above, the present author has sug- 
gested that the speakers of Proto-Armenian arrived in 
the Upper Euphrates and Arsanias (Muratsu) valleys 
in the 12th century B.C. under the name of /*Mus-k c i/. 
If that were the case, borrowings from Proto-Armenian 
into Urartian (texts from the 8th to the 6th cen- 
tury B.C.!) would be possible. I have sorted out three 
glosses, none of which are attested in Hurrian (all 
Hurrian texts and glosses known to us antedate the 
12th century B.C.!). Two of them have plausible PEC 
etymologies, but etymologies from Proto-Armenian 
seem at least equally plausible. The following list 
should be regarded as provisional and experimental 
and is intended to invite discussion. 

1. Urart. Arsiha /arciva/ name of a horse, pre- 
sumably 'Eagle'. The proposed PEC etymology is from 

phonotactic situation. At the same time a phoneme c or c, 
seems to have existed in Urartian, cf. Diakonoff and Starostin, 
op. cit. 

* c warcciw c V, attested as the etymon for the name of 
the mountain eagle in several Eastern Caucasian 
languages; the development to Urart. /arcivs/ is 
regular. The alternative IE etymology is < *rg'i-p[-o-, 
OInd rjipya-, A vest, drdzifya- 'eagle'; less convincing is 
the connection with Greek dpyuTUog 'kite'. 

The Armenian equivalent is arcui, var. arciw, hence 
also, according to A. G. Sanidze, the Georgian arcivi. 

It seems improbable that the mountaineers should 
import the name of the mountain eagle, and the 
Caucasian reflexes of * c w arcciw c Fare reliable enough; 
it is probable that Urartian, an Eastern Caucasian 
language, should have a PEC name for the eagle, and 
that the Armenian (and Georgian) word should be 
derived from Urartian. Moreover, the Indo-Iranian 
rjipya- 1 drazifya- could have been borrowed from the 
mountaineers during the sojourn of the Indo-Iranian 
speakers near the Caucasus: the etymology in IE might 
have been a Volksetymologie. Moreover, the Indo- 
Europeans actually had their own name for the 
(steppe-)eagle, *Har-. Nevertheless, a borrowing from 
Proto-Armenian into Urartian is not beynd the bounds 
of possibility, and Arm. arcui is easily derived from an 
IE *rg'i-pi-o-. This, however, would mean that the 
change of the IE intervocal *-p- to -w-, which Kortlandt 
places at his stage 10 (*/;>*(p) and 13 (*-(p- > -w-) 
antedated the contact with Hurro-Urartian speakers. 

2. Urartian burg-ana- (read /burg-, borg-, purg-, porg-, 
purg-, porg-ana-/), 'column (?)'. We had proposed 
a PEC etymology, but it was based on the assumption 
that the word meant 'tower'. This etymology should 
now be abandoned. It had also been assumed that 
Arm. hurgn 'tower, pyramid' is a borrowing from 
Urart. burg-ana- allegedly 'tower'. However, (1) the 
word is not attested in Hurrian and therefore cannot be 
proved to be originally Hurro-Urartian; (2) the 
Armenian reflex of burg-ana- would have been, as 
pointed out to me by Perikhanian, not burgn but 
*brgan; (3) as shown by Ch. de Lamberterie, 26 burgn 
'tower' holds a place in the system of Armenian 
reflexes of the root IE *bhrg'h-/*bh[gh-, similarly to 
durgn 'potter's wheel' from IE *dhrg r h-/*dhrgh-: darj- 
'I (re)turned', darnam 'I (re)turn', durgn, gen. drgan 
'potter's wheel'; bar/- 'I lifted', barnam 'I lift', -berj 
'height', burgn 'tower, pyramid'; (4) Urart. burg-ana- 
apparently does not mean 'tower'. 

Now it is hard to imagine that the Hurro-Urartians 
should have to borrow the word for 'tower' from 
Proto-Armenian, since dwelling- and battle-towers 
had been known in the Hurro-Urartian region for 

Ch. de Lamberterie, "Echange de gutturales en armenien," 
Annual of Armenian linguistics 1 (1980), 27-28. 

Diakonoff: Hurra- Urartian Borrowings in Old Armenian 


millennia. However, a word for some specific vertical 
object, like a stela or a column might have been 
borrowed. 27 

The problem is still more complicated by the fact of 
the existence of an Urart. burg-ala- which apparently 
means 'ally'. 

3. Urart. ulgusd (to read /ulg-, ulq-, uly-, olg-, 
olq- or oly-o-sa/) expressed by the Sumero-Akkadian 
logogram for balatu 'life' but more probably denoting 
'health, well-being, the being alive' (paralleling 
/pic-o-sa/ 'rejoicing'). This word (or, better, its stem 

Syrian burg-a and Arab, bur]- have no connection with 
Arm. burgn: they are derived from Late Latin burgus, a 
borrowing from Germanic — which incidentally shows that a 
word for 'tower, fortress' can also be borrowed by the 
language of a nation long familiar with all kinds of fortifica- 

*ulg-, since /-o-sa/ are abstract word-forming suffixes) 
has been proposed as an etymon for Arm. off 'whole, 
alive'. As pointed out to me by Perikhanian, this 
etymology is wrong, because Arm. olj is < IE *ol-[o- 
'whole' (cf. Arm. sterj 'barren' < IE *ster-io-, Greek 
oxeipa 'barren', Lat. ster-ilis, etc.). 28 

However, since *ulg- has so far not been attested in 
Hurrian (nor in Urartian, except for the word ulgusd 
itself), it is permissible to etymologize Urart. ulgusd 
(reading it /oly-o-sa/) from a Proto-Armenian *olyo-. 29 

On this group of words see Kortlandt, op. cit., p. 104. 

I am very grateful to Professor Anahit Perikhanian for 
help in identifying the strata of borrowings (especially the two 
strata of Aramaic borrowings), for checking the Armenian 
glosses, for bringing to my notice a number of Armenological 
studies, and for consultation on matters of Indo-European